Plotting candlestick data in Python using Webull's API
To import and login Webull (plugin your email and account password):
!pip install webull
from webull import webull
wb = webull()
wb.login('my@email.com', 'password')
To obtain stock data, we'll use the get_bars function and specify the stock's symbol, timeframe, and number of samples (i.e. stock, interval, and count) and save it as a DataFrame using pandas's to_csv function:
stock_symbol = 'SPY'
stock_data = wb.get_bars(stock=stock_symbol, interval='m1', count=390, extendTrading=0)
_date = '2023-09-26'
file_name = f'/content/drive/My Drive/Colab Notebooks/DATA_FOLDERS/DATA_FRAMES/{stock_symbol}_{_date}.csv'
import pandas as pd
stock_data.to_csv(file_name)
where 'm1' refers to a one-minute timeframe.
We'll then retrieve our saved stock data using pandas's read_csv function and parse the timestamps into time and date:
df = pd.read_csv(file_name)
def parse_date_from_timestamp(timestamp):
return timestamp[:timestamp.find(' ')]
def parse_time_from_timestamp(timestamp):
def correct_timestamps(timestamp):
'''convert timestamp from 24h to 12h'''
return timestamp if int(timestamp[:2]) <= 12 else '0' + str(int(timestamp[:2]) - 12) + timestamp[2:]
return correct_timestamps(timestamp[timestamp.find(' ') + 1:timestamp.find('-4:') - 5])
df['date'] = df.timestamp.map(parse_date_from_timestamp)
df['timestamp'] = df.timestamp.map(parse_time_from_timestamp)
We'll define a candlestick_plot_function to group the data into numpy arrays and then plot the data as candlesticks using a combination of BoxStyle, FancyBboxPatch, and Line2D from matplotlib:
import numpy as np
from matplotlib import pyplot as plt
from matplotlib.patches import BoxStyle
from matplotlib.patches import FancyBboxPatch
from matplotlib.lines import Line2D
def candlestick_plot_function(fig, ax, df, stock_symbol, candlestick_size_in_minutes=30, wick_linewidth=2.0, fancy_box_padding=0.0005):
def calc_num_candlesticks(df, candlestick_size_in_minutes):
return int(df.index.stop / candlestick_size_in_minutes) if df.index.stop / candlestick_size_in_minutes == int(df.index.stop / candlestick_size_in_minutes) else int(df.index.stop / candlestick_size_in_minutes) + 1
def calc_box_width(num_candlesticks):
return 0.98 - (num_candlesticks / 125)
def remove_prefixed_zero(timestamp):
return timestamp[:5] if timestamp[0] != '0' else timestamp[1:5]
y = np.zeros((calc_num_candlesticks(df, candlestick_size_in_minutes), 4)).astype(float)
box_width = calc_box_width(y.shape[0])
ax.set_facecolor([0, 0, 0.35])
ax.grid(which='major', axis='both', color=[1, 1, 1], linewidth=0.5, zorder=0)
for candlestick in range(y.shape[0]):
indexes_per_candlestick = range(candlestick*candlestick_size_in_minutes, (candlestick + 1)*candlestick_size_in_minutes, 1)
if indexes_per_candlestick.stop > df.index.stop:
indexes_per_candlestick = range(candlestick*candlestick_size_in_minutes, df.index.stop, 1)
data = df.iloc[indexes_per_candlestick]
y[candlestick, :] = np.array([data.open.iloc[0], np.max(data.high), np.min(data.low), data.close.iloc[-1]])
top_of_box = np.max([y[candlestick, 0], y[candlestick, 3]])
bottom_of_box = np.min([y[candlestick, 0], y[candlestick, 3]])
box_color = np.array([0.0, 0.8, 0.6941]) if y[candlestick, 0] < y[candlestick, 3] else np.array([1.0, 0.0, 0.0])
ax.add_line(Line2D(xdata=(candlestick, candlestick), ydata=(y[candlestick, 2], bottom_of_box), color=box_color, linewidth=wick_linewidth, antialiased=True, zorder=2))
ax.add_line(Line2D(xdata=(candlestick, candlestick), ydata=(y[candlestick, 1], bottom_of_box), color=box_color, linewidth=wick_linewidth, antialiased=True, zorder=2))
ax.add_patch(FancyBboxPatch(xy=(candlestick - box_width*0.5, bottom_of_box), width=box_width, height=top_of_box - bottom_of_box, facecolor=box_color, edgecolor=box_color, boxstyle=BoxStyle('round', pad=fancy_box_padding), zorder=2))
ax.set_ylim(np.min(y) - 0.1*(np.max(y) - np.min(y)), np.max(y) + 0.1*(np.max(y) - np.min(y)))
ax.set_xlim(-0.1*y.shape[0], y.shape[0] + 0.1*y.shape[0])
ax.set_title(f'''
{stock_symbol}
{candlestick_size_in_minutes} Minute Candlesticks''', fontsize=15, fontweight='bold')
ax.set_yticks([_y for _y in ax.get_yticks()][1:-1])
ax.set_yticklabels(['{:.2f}'.format(_y) for _y in ax.get_yticks()])
ax.set_xticks([0, int(y.shape[0]*0.5), y.shape[0] - 1])
ax.set_xticklabels([remove_prefixed_zero(df.timestamp.values[0]), remove_prefixed_zero(df.timestamp.values[int(df.timestamp.values.shape[0]*0.5)]), remove_prefixed_zero(df.timestamp.values[-1])])
for axis in ['left', 'right', 'top', 'bottom']:
ax.spines[axis].set_visible(False) if axis in ['top', 'right'] else ax.spines[axis].set_linewidth(5)
for label in (ax.get_xticklabels() + ax.get_yticklabels()):
label.set_fontsize(15)
label.set_fontweight('bold')
fig = plt.figure(figsize=(5, 5))
ax = plt.subplot(1, 1, 1)
candlestick_plot_function(fig, ax, df, stock_symbol)
plt.show()
Which creates a 30-minute candlesticks plot:
We can also adjust the candlestick_size_in_minutes parameter to plot different timeframes; for example 10-minute candlesticks:
fig = plt.figure(figsize=(5, 5))
ax = plt.subplot(1, 1, 1)
candlestick_plot_function(fig, ax, df, stock_symbol, candlestick_size_in_minutes=10, wick_linewidth=1.0)
plt.show()
or 60-minute candlesticks:
fig = plt.figure(figsize=(5, 5))
ax = plt.subplot(1, 1, 1)
candlestick_plot_function(fig, ax, df, stock_symbol, candlestick_size_in_minutes=60)
plt.show()
First we need a list of start and stop times as hour and minute integers that can span across one or more days:
def get_candlesticks(df, candlestick_size_in_minutes=30):
'''returns a list of pairs of time stamps as integers [hour (start), minute (start), next hour (stop), next minute (stop)] from 9:30am up to (but not including) 4:00pm grouped by the candlestick_size_in_minutes parameter'''
__hour = df.hour.iloc[0]
def round_minute_down(__minute, candlestick_size_in_minutes):
candlesticks = np.arange(0, 60 + candlestick_size_in_minutes, candlestick_size_in_minutes).astype(int)
first_candlestick = np.where(np.abs(candlesticks - __minute) == np.min(np.abs(candlesticks - __minute)))
if candlesticks[first_candlestick] > __minute:
first_candlestick[0][0] -= 1
return candlesticks[first_candlestick[0][0]]
__minute = round_minute_down(df.minute.iloc[0], candlestick_size_in_minutes)
__date = 0
candlesticks = [[__hour, __minute, __hour + 1 if __minute + candlestick_size_in_minutes > 60 else __hour, min(__minute + candlestick_size_in_minutes, 60), np.unique(np.array(df.date))[__date]]]
for _ in range(int((390 - candlestick_size_in_minutes) / candlestick_size_in_minutes)*np.unique(np.array(df.date)).shape[0]):
__minute += candlestick_size_in_minutes
if __minute >= 60:
__minute -= 60
__hour += 1
if __hour == 13:
__hour = 1
if __hour == 4:
__hour = 9
__minute = 30
__date += 1
if __date >= np.unique(np.array(df.date)).shape[0] - 1 and candlesticks[-1][1] <= df.minute.iloc[-1] < candlesticks[-1][3] and candlesticks[-1][0] == df.hour.iloc[-1] and candlesticks[-1][4] == df.date.iloc[-1]:
return candlesticks
candlesticks.append([__hour, __minute, __hour + 1 if __minute + candlestick_size_in_minutes > 60 else __hour, min(__minute + candlestick_size_in_minutes, 60), df.date.iloc[0] if np.unique(np.array(df.date)).shape[0] == 1 else np.unique(np.array(df.date))[min(__date, np.unique(np.array(df.date)).shape[0] - 1)]])
return candlesticks
We'll then define a function that returns True for indexes that fall between the start and stop hours and minutes list and match the correct 'date':
def is_a_candlestick(df_hours, df_minutes, _hour, _minute, _next_hour, _next_minute, df_date, _date):
'''returns True for timestamps ([hour, minute]) that match the correct candlestick; conditions vectorized for numpy.where() function'''
if all([df_date == _date, df_hours == _hour, _hour == _next_hour, df_minutes >= _minute, df_minutes < _next_minute]) or all([df_date == _date, df_hours == _hour, _hour != _next_hour, df_minutes >= _minute, df_minutes < _next_minute if _minute < _next_minute else True]) or all([df_date == _date, df_hours == _next_hour, _next_minute != 60, df_minutes < _next_minute, df_minutes >= _minute if _minute < _next_minute else True]):
return True
is_a_candlestick = np.vectorize(is_a_candlestick)
What numpy's vectorize function does is it allows us to define a function hook into which another numpy function can get as a vectorized input. In the simplest case, which we did, is use an if-statement that returns True, which when passed to numpy's where() function, will provide us the indexes that meet our condition(s), without having to iterate through each one:
for candlestick, (_hour, _minute, _next_hour, _next_minute, _date) in enumerate(candlesticks):
indexes = np.where(is_a_candlestick(np.array(df.hour), np.array(df.minute), _hour, _minute, _next_hour, _next_minute, np.array(df.date), _date))
y[candlestick, :] = np.array([df.open.iloc[indexes].iloc[0], np.max(df.high.iloc[indexes]), np.min(df.low.iloc[indexes]), df.close.iloc[indexes].iloc[-1]])
Altogether we have:
import numpy as np
from matplotlib import pyplot as plt
from matplotlib.patches import BoxStyle
from matplotlib.patches import FancyBboxPatch
from matplotlib.lines import Line2D
def candlestick_plot_function(fig, ax, df, stock_symbol, candlestick_size_in_minutes=30, wick_linewidth=2.0, fancy_box_padding=0.0005):
def get_candlesticks(df, candlestick_size_in_minutes=30):
'''returns a list of pairs of time stamps as integers [hour (start), minute (start), next hour (stop), next minute (stop)] from 9:30am up to (but not including) 4:00pm grouped by the candlestick_size_in_minutes parameter'''
__hour = df.hour.iloc[0]
def round_minute_down(__minute, candlestick_size_in_minutes):
candlesticks = np.arange(0, 60 + candlestick_size_in_minutes, candlestick_size_in_minutes).astype(int)
first_candlestick = np.where(np.abs(candlesticks - __minute) == np.min(np.abs(candlesticks - __minute)))
if candlesticks[first_candlestick] > __minute:
first_candlestick[0][0] -= 1
return candlesticks[first_candlestick[0][0]]
__minute = round_minute_down(df.minute.iloc[0], candlestick_size_in_minutes)
__date = 0
candlesticks = [[__hour, __minute, __hour + 1 if __minute + candlestick_size_in_minutes > 60 else __hour, min(__minute + candlestick_size_in_minutes, 60), np.unique(np.array(df.date))[__date]]]
for _ in range(int((390 - candlestick_size_in_minutes) / candlestick_size_in_minutes)*np.unique(np.array(df.date)).shape[0]):
__minute += candlestick_size_in_minutes
if __minute >= 60:
__minute -= 60
__hour += 1
if __hour == 13:
__hour = 1
if __hour == 4:
__hour = 9
__minute = 30
__date += 1
if __date >= np.unique(np.array(df.date)).shape[0] - 1 and candlesticks[-1][1] <= df.minute.iloc[-1] < candlesticks[-1][3] and candlesticks[-1][0] == df.hour.iloc[-1] and candlesticks[-1][4] == df.date.iloc[-1]:
return candlesticks
candlesticks.append([__hour, __minute, __hour + 1 if __minute + candlestick_size_in_minutes > 60 else __hour, min(__minute + candlestick_size_in_minutes, 60), df.date.iloc[0] if np.unique(np.array(df.date)).shape[0] == 1 else np.unique(np.array(df.date))[min(__date, np.unique(np.array(df.date)).shape[0] - 1)]])
return candlesticks
def calc_box_width(num_candlesticks):
'''returns a box_width value such that candlesticks are close together but not touching or overlapping'''
return 0.98 - (num_candlesticks / 125)
def remove_prefixed_zero(timestamp):
'''returns for example "9:30" in place of "09:30"'''
return timestamp[:5] if timestamp[0] != '0' else timestamp[1:5]
def is_a_candlestick(df_hours, df_minutes, _hour, _minute, _next_hour, _next_minute, df_date, _date):
'''returns True for timestamps ([hour, minute]) that match the correct candlestick; conditions vectorized for numpy.where() function'''
if all([df_date == _date, df_hours == _hour, _hour == _next_hour, df_minutes >= _minute, df_minutes < _next_minute]) or all([df_date == _date, df_hours == _hour, _hour != _next_hour, df_minutes >= _minute, df_minutes < _next_minute if _minute < _next_minute else True]) or all([df_date == _date, df_hours == _next_hour, _next_minute != 60, df_minutes < _next_minute, df_minutes >= _minute if _minute < _next_minute else True]):
return True
is_a_candlestick = np.vectorize(is_a_candlestick)
candlesticks = get_candlesticks(df, candlestick_size_in_minutes)
y = np.zeros((len(candlesticks), 4)).astype(float)
box_width = calc_box_width(y.shape[0])
ax.set_facecolor([0, 0, 0.35])
ax.grid(which='major', axis='both', color=[1, 1, 1], linewidth=0.5, zorder=0)
for candlestick, (_hour, _minute, _next_hour, _next_minute, _date) in enumerate(candlesticks):
indexes = np.where(is_a_candlestick(np.array(df.hour), np.array(df.minute), _hour, _minute, _next_hour, _next_minute, np.array(df.date), _date))
y[candlestick, :] = np.array([df.open.iloc[indexes].iloc[0], np.max(df.high.iloc[indexes]), np.min(df.low.iloc[indexes]), df.close.iloc[indexes].iloc[-1]])
top_of_box = np.max([y[candlestick, 0], y[candlestick, 3]])
bottom_of_box = np.min([y[candlestick, 0], y[candlestick, 3]])
box_color = np.array([0.0, 0.8, 0.6941]) if y[candlestick, 0] < y[candlestick, 3] else np.array([1.0, 0.0, 0.0])
ax.add_line(Line2D(xdata=(candlestick, candlestick), ydata=(y[candlestick, 2], bottom_of_box), color=box_color, linewidth=wick_linewidth, antialiased=True, zorder=2))
ax.add_line(Line2D(xdata=(candlestick, candlestick), ydata=(y[candlestick, 1], bottom_of_box), color=box_color, linewidth=wick_linewidth, antialiased=True, zorder=2))
ax.add_patch(FancyBboxPatch(xy=(candlestick - box_width*0.5, bottom_of_box), width=box_width, height=top_of_box - bottom_of_box, facecolor=box_color, edgecolor=box_color, boxstyle=BoxStyle('round', pad=fancy_box_padding), zorder=2))
ax.set_ylim(np.min(y) - 0.1*(np.max(y) - np.min(y)), np.max(y) + 0.1*(np.max(y) - np.min(y)))
ax.set_xlim(-0.1*y.shape[0], y.shape[0] + 0.1*y.shape[0])
ax.set_title(f'''
{stock_symbol}
{candlestick_size_in_minutes} Minute Candlesticks''', fontsize=15, fontweight='bold')
ax.set_yticks([_y for _y in ax.get_yticks()][1:-1])
ax.set_yticklabels(['{:.2f}'.format(_y) for _y in ax.get_yticks()])
ax.set_xticks([0, int(y.shape[0]*0.5), y.shape[0] - 1])
ax.set_xticklabels([remove_prefixed_zero(df.timestamp.values[0]), remove_prefixed_zero(df.timestamp.values[int(df.timestamp.values.shape[0]*0.5)]), remove_prefixed_zero(df.timestamp.values[-1])])
for axis in ['left', 'right', 'top', 'bottom']:
ax.spines[axis].set_visible(False) if axis in ['top', 'right'] else ax.spines[axis].set_linewidth(5)
for label in (ax.get_xticklabels() + ax.get_yticklabels()):
label.set_fontsize(15)
label.set_fontweight('bold')
stock_symbol = 'SPY'
_date = '2023-09-27'
file_name = f'/content/drive/My Drive/Colab Notebooks/DATA_FOLDERS/DATA_FRAMES/{stock_symbol}_{_date}.csv'
import pandas as pd
df = pd.read_csv(file_name)
def parse_date_from_timestamp(timestamp):
return timestamp[:timestamp.find(' ')]
def parse_time_from_timestamp(timestamp):
def correct_timestamps(timestamp):
'''convert timestamp from 24h to 12h'''
return timestamp if int(timestamp[:2]) <= 12 else '0' + str(int(timestamp[:2]) - 12) + timestamp[2:]
return correct_timestamps(timestamp[timestamp.find(' ') + 1:timestamp.find('-4:') - 5])
df['date'] = df.timestamp.map(parse_date_from_timestamp)
df['timestamp'] = df.timestamp.map(parse_time_from_timestamp)
def get_hour(timestamp):
return int(timestamp[:2])
def get_minute(timestamp):
return int(timestamp[3:5])
df['hour'] = df.timestamp.map(get_hour)
df['minute'] = df.timestamp.map(get_minute)
fig = plt.figure(figsize=(5, 5))
ax = plt.subplot(1, 1, 1)
candlestick_plot_function(fig, ax, df, stock_symbol, candlestick_size_in_minutes=30)
plt.show()
fig = plt.figure(figsize=(5, 5))
ax = plt.subplot(1, 1, 1)
candlestick_plot_function(fig, ax, df, stock_symbol, candlestick_size_in_minutes=60)
plt.show()
